The growth of social network sites such as MySpace and Facebook has been one of the most interesting features of this century's web, and it's attracting a lot of academic interest. There's now a good overview of the field, Social Network Sites: Definition, History, and Scholarship, by Danah Boyd and Nicole Ellison in the Journal of Computer-Mediated Communication.
One thing that's well worth a look is the timeline diagram (above), from Six Degrees on. Another interesting aspect is that different social network sites dominate in different geographical areas. As the paper says:
Friendster gained traction in the Pacific Islands, Orkut became the premier SNS in Brazil before growing rapidly in India, Mixi attained widespread adoption in Japan, LunarStorm took off in Sweden, Dutch users embraced Hyves, Grono captured Poland, Hi5 was adopted in smaller countries in Latin America, South America, and Europe, and Bebo became very popular in the United Kingdom, New Zealand, and Australia.
There's a cascade effect, that can be more powerful than either features or quality. Clearly it makes sense to join the same network as most of your friends, so each local market will tend to tip towards whichever one that is. However, while the initial decision made by early adopters may be influenced by a particular feature (for example, Hi5's Spanish language version), it may also have a large random element.
In the long run, global systems tend to displace local ones, so it will be interesting to see what the market looks like in another five or 10 years.
